谱修正迭代结果的协因数矩阵

The Co-factor Matrix of the Iteration Method by Correcting Characteristic Value

导出了谱修正迭代结果的协因数矩阵Q^X^X ,证明了当法方程系数矩阵N满秩且呈良态时 ,Q^X^X 就是N的凯利逆N- 1 ;当N秩亏时 ,Q^X^X 就是N的Moore Penrose逆N+ 。

In the parameter estimation, there exist two problems: changing the equivalences of the equations makes estimation results biased; determining the ridge parameter k is very difficult and random. To solve the two problems, this paper presents the iteration method by correcting characteristic value. The co-factor matrix Q is derived firstly, then proves that Q is just the Kaley inverse N -1 of N in the case of rank-full, which is Moore-Penrose inverse N + of N in the case of rank-defect.